UCLA

A primary function of prosody in many languages is to convey information structure--the "packaging" of a sentence's content into categories such as "focus", "given" and "topic". In English and other West Germanic languages it is widely assumed that focus is signaled prosodically by the location of a nuclear pitch accent. As a result, prenuclear, or "secondary" accents are standardly regarded as optional, phonological objects that are unrelated to the information structural representation. This dissertation investigates, from the perspective of the listener, how valid this claim about prenuclear accents is.

As a case study, I consider a putative prosodic ambiguity: the size of the focus constituent in English SVO constructions (i.e., "broad focus" on a VP versus "narrow focus" on an object). My approach to this issue is essentially a three-pronged one, considering the production, perception and processing of prenuclear accents in relation to this contrast. Recent phonetic evidence from production studies is the starting point for a set of perception experiments (Chapter 2) and a pair of cross-modal priming experiments (Chapter 3). Both sets of experiments provide evidence that listeners have expectations about focus-prenuclear accent correspondences that mirror patterns reported in speakers' productions, suggesting that the broad versus narrow focus contrast is not a genuine prosodic ambiguity. An additional matter that is investigated is the extent to which individual differences in "cognitive processing styles" (autistic traits and verbal working memory) contribute to variation among listeners.

To account for the experimental findings, I argue that the prosodic realization of the size of the focus constituent in English SVOs represents conventionalized, phonological behavior. The variation, it is shown, can be captured by an Autosegmental Metrical model of prosodic structure that includes syntagmatic relations of tonal prominence along the lines proposed by Ladd (1990). This level of "tonal metrical structure" represents linguistically-specified pitch range, and it is demonstrated that such structure is independently needed to account for the prosodic realization of several other information structural contrasts.